Mercurial > repos > matt-shirley > sra_tools

annotate sra.py @ 1:28fda7631857 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
minor fixes
author Matt Shirley <mdshw5@gmail.com>
date Wed, 29 May 2013 22:08:19 -0400
parents ffdd41766195
children b77840618b8f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
1 """
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
2 Sra class
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
3 """
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
4
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
5 import galaxy.datatypes.binary
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
6 from galaxy.datatypes.binary import Binary
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
7 import data, logging, binascii
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
8 from galaxy.datatypes.metadata import MetadataElement
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
9 from galaxy.datatypes import metadata
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
10 from galaxy.datatypes.sniff import *
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
11 from galaxy import eggs
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
12 import pkg_resources
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
13 pkg_resources.require( 'bx-python' )
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
14 import os, subprocess, tempfile
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
15 import struct
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
16
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
17 class Sra( Binary ):
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
18 """ Sequence Read Archive (SRA) """
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
19 file_ext = 'sra'
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
20
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
21 def __init__( self, **kwd ):
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
22 Binary.__init__( self, **kwd )
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
23 def sniff( self, filename ):
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
24 """ The first 8 bytes of any NCBI sra file is 'NCIB.sra', and the file is binary. Not sure if EBI and DDBJ files may differ.
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
25 For details about the format, see http://www.ncbi.nlm.nih.gov/books/n/helpsra/SRA_Overview_BK/#SRA_Overview_BK.4_SRA_Data_Structure
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
26 """
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
27 try:
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
28 header = open(filename).read(8)
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
29 if binascii.b2a_hex(header) == binascii.hexlify('NCBI.sra'):
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
30 return True
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
31 return False
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
32 except:
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
33 return False
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
34 def set_peek(self, dataset, is_multi_byte=False):
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
35 if not dataset.dataset.purged:
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
36 dataset.peek = 'Binary sra file'
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
37 dataset.blurb = data.nice_size(dataset.get_size())
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
38 else:
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
39 dataset.peek = 'file does not exist'
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
40 dataset.blurb = 'file purged from disk'
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
41 def display_peek(self, dataset):
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
42 try:
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
43 return dataset.peek
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
44 except:
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
45 return 'Binary sra file (%s)' % ( data.nice_size(dataset.get_size()))
0
ffdd41766195 Initial version - still need to test if datatype works correctly, and implement scripted download of SRA binaries.
matt-shirley <mdshw5@gmail.com>
parents:
diff changeset
46
1
28fda7631857 minor fixes
Matt Shirley <mdshw5@gmail.com>
parents: 0
diff changeset
47 Binary.register_sniffable_binary_format('sra', 'sra', Sra)